Suggested Standard Operating Procedures
We present this summary of the SOP with links to sections of the SOP. In this summary, we assume that the user has a split-beam unit, likely 120 kHz Biosonics or Simrad. We also assume the user has access to a software package, likely Echoview or Sonar5.
In preparation for the survey, the following steps should be taken:
- Chose a deployment method (Equipment deployment).
- Chose a survey design (Survey design).
- Calibrate the echosounder in both SV and TS domain with settings used during the survey (pulse duration, power settings) (System calibration).
- Test the equipment during standard operating speed and use of other onboard equipment (e.g., trawl winches). Attempt to minimize noise (Noise).
- Record passive data during standard survey speed (Passive data).
- Consider collecting stationary data to test for bubbles and get long traces on single fish (Stationary sounding).
During data collection, there are a number of collection settings that cannot be changed later during analysis. These are listed here together with suggested values:
- Collect raw data to below the bottom or maximum range of usable data (Collection settings).
- Set pulse length to 0.4 ms (0.2 to 0.6 ms acceptable, 0.256 or 0.512 ms on Simrad) (Collection settings).
- Set power to 300 W or lower for Simrad, default for Biosonics (System settings). If operating at higher frequencies than 120 kHz, use lower power settings.
- Set the ping interval slow enough to avoid shadow bottom (0.5-4 pings•s-1) (Collection settings).
- Set the collection threshold to -100 dB or lower (squared threshold in Biosonics) (Collection settings).
- There are a number of other collection settings for Simrad and a few more for Biosonics, but they are not as critical during data acquisition as they can be changed during post-processing after raw data is collected (Correction of incorrect settings).
Standard operating procedures for data analysis (also referred to as post-processing) are listed next. The order of implementation varies somewhat among analysis programs. Statements related to EchoView (SonarData 2006) are designated with EV, statements related to Sonar5 (Balk and Lindem 2007) with S5.
- Enter sound speed, absorption coefficient, and system calibration settings (System and environmental settings).
Calculate average sound speed and absorption coefficient using the software calculator given the depth of the fish of interest and the measured temperature gradient.
- Adjust transducer depth.
Add depth of transducer so all depths are relative to the surface.
- Synchronize time.
If multiple echosounders are used, make sure the time and depths are synchronized. Time synchronization can be achieved by aligning a distinguishable point on the echograms from both units.
- Add a surface exclusion line (Surface exclusion zone.).
Add a surface line at a depth of at least twice the near-field and exclude data above that line. Larger surface exclusion zones may be needed if there is a lot of surface noise. Note that the surface line should encompass the near field and the depth of the transducer.
- Detect and correct the bottom (Bottom exclusion zone.).
Run bottom detection algorithm of your software using default parameters. The bottom line should be set slightly higher than detected depth to avoid including data from the bottom dead zone and poorly defined bottoms. We suggest 0.5 m (0.2 to 1 m acceptable). Inspect all data for accurate bottom detection and redefine as needed.
- Scrutinize echograms for bad data and remove such regions from the analysis (Noise removal.)
Mark (EV) or erase/threshold (S5) bad data regions and exclude from the analysis. Note that the assumption about fish density in bad data regions is important if these regions are large.
- Remove ambient noise. (Noise removal.)
Calculate noise at 1 m depth in the SV domain. Remove noise by applying the noise subtraction algorithm (S5) or by integrating noise levels and subtracting after data export (EV). Noise at 1 m is different for TS and SV data, but can be calculated from each other.
- Extract TS data from single echo detection. (Single echo detection.)
Run single echo detection algorithms. Use the same method as used when measuring TS of the calibration ball (EV). This step is done in the initial conversion step in S5. Suggested initial settings are -75 dB for lower threshold, 0.6 and 1.5 for minimum and maximum pulse duration multiplier, determination of echo length at -6dB from peak value, 6 dB for maximum beam compensation (2-way (EV), 3dB for 1-way (S5)), variance for phase jitter of 0.6 in both directions (mechanical degrees).
- Study the TS distribution at different depths. (Separating groups of interest.)
Use the TS versus Depth graph. Note that changes in TS distribution with depth may indicate a change in fish species or age group composition with depth or a bias related to the sampling volume and number of targets. Compare different regions of the lake (nearshore, offshore). Define depth layers for analysis based on this TS graph. The data should be analyzed in depth layers with as homogeneous fish species and size structure as possible. Different depth regions may have to be used in different parts of the lake.
- Set the minimum TS of interest. (Separating groups of interest.).
Based on the measured TS distribution and/or known TS distribution of the fish species of interest, set the minimum TS. Check the TS distribution to -75dB to see if smaller targets are likely to be included in the fish TS range. Common values for minimum TS of interest range from -65 to -55 dB, but this depend on the situation.
- Set the integration threshold. (Separating groups of interest.)
The integration threshold should be set to 6 dB below the minimum TS threshold (in the TS domain) to account for observations of all targets above the minimum TS within the half power beam width. Set this threshold in the Amp echogram representing 40 log R data (S5) or use the constant TS threshold in the SV echogram (EV). Include any difference between calibration in SV and TS (EV).
- Choose the size of the analysis cell. (Analysis cell size.)
This should include 20 to 50 single fish echoes in most cells. We suggest 200 to 500 m long horizontal intervals/segments. Range (depth) layers need to be chosen given the fish distributions (see step 10). Integration intervals/segments could be shorter if using geostatistics and fish are abundant, or longer if few fish are present. If the depth layers are small, the horizontal distance may need to be increased to get sufficient number of targets in the analysis cells.
- Export integration and TS data. (File exports.)
SV and ABC are obtained from the Amp echogram (S5) or the converted SV echogram (EV, which has a TS based threshold applied – e.g. -66dB). TS distribution and mean σbs are obtained from the single target detection echogram (EV) or SED echogram (S5). Apply the lower TS of interest (e.g. -60 dB) as the lower data threshold for TS data. EV requires 3 separate export runs (for SV, mean TS, and TS distribution). Merge these data in a database. S5 exports all information in several table formats.
- Check for bias in in situ TS. (Backscattering.)
Check for quality of in situ TS measures for each analysis cell using the NV index. Use the average σbs by depth region to calculate the NV index. This is used to test if in situ TS can be used for density estimates. If NV is >0.1, replace the average σbs in that cell with average σbs in surrounding cells or from the appropriate depth layer.
- Calculate fish density. (Density, Abundance.)
Calculate fish density by dividing ABC with mean σbs for each analysis cell. A density per unit area is obtained by summing over all layers in each interval/segment. Fish/ha is obtained from fish•m-2 by multiplying by 10,000. Be careful if using NASC to include the 4π term (Table 1).
- Apportion the acoustic fish density to different fish species. (Species specific abundance.)
This is a critical step that should rely on current and previous experience from the lake, temperature profiles, and catch data. Catch data can be incorporated in S5 to directly export density by fish species. TS distributions can also be used.
- Make assumptions about fish in surface and bottom zones. (Species specific abundance.)
Decide how to deal with fish present in the surface and bottom exclusion zones (4.9). The report should be specific on what assumptions are made for these acoustic dead zones and if the choice was to exclude these areas from analysis.
- Calculate fish density by species. (Abundance, Species specific abundance.)
Calculate average fish density by species with the appropriate statistics for the survey design used.
- Calculate uncertainty. (Uncertainty.)
Calculate uncertainty of the estimate including all aspects of uncertainty that is known at the time. List the sources of uncertainty included in these calculations.
It is a good idea to check for any unreasonable fish densities. Refer back to the echogram to see if the segments with very high gish densities match up with dense fish areas on the echograms.
Reports presenting fish density derived from acoustic data should include the following information:
- Hardware and software used, including version.
- Ping rates, pulse length, field calibration information, beam width.
- Single echo detection parameters.
- The minimum threshold level that is considered to represent the fish of interest and the method used for noise removal.
- The noise level at 1 m (SV domain preferred) and the detection limit (range) for the smallest fish of interest.
- Information on the number of analysis cells with high Nv values.
- A graph of representative TS distributions for layers with differences in this distribution. Alternatively, provide a graph of TS versus depth.
- Information on decision rules for allocation of fish density to different species.
- Average fish density and method used for calculations (geostatistics, cluster sampling, etc.).
- Estimates of uncertainty, including list the factors included in the estimates.
- Map of the spatial distribution of fish density along transects to visualize spatial patterns and variability.